一种基于层次分析法的视频摘要评价模型

视频摘要是对视频内容进行浓缩的一项技术，对于快速了解视频内容至关重要。如何对视频摘要的效果进行评价，是值得研究的一个问题。论文基于层次分析法构建了视频摘要评价模型，将视频摘要质量作为最终评价目标，以内容合理性和结构合理性作为准则，以内容完整性、特殊重要性、整体流畅性等作为测度层，从而建立了视频摘要评价指标体系。最后，通过对随机生成、基于特写人脸的摘要生成以及融合视音频特征的摘要生成三种算法对所提评价方法进行了实验验证，表明该方法能够有效反映出视频摘要的质量。

An Evaluation Model of Video Summarization Based on Analytic Hierarchy Process

The goal of video summarization is to summary the video content, which is important to help to grasp the main idea of video content. How to evaluate the summarization effect is a problem needed urgently to be solved. In this paper, it constructs a video summarization evaluation model based on Analytic Hierarchy Process (AHP) method. In this model, the target is the quality of video summarization, the rule is the content and structure rationality, and the measurement layer is the integrality, the importance and the fluency of video content. These construct the evaluation factor system of video summarization. Finally, by experiments on three different algorithms such as random generation, feature face-based generation and audio-visual feature fusion generation, the proposed method is proved to be effective in reflecting the quality of video content.